Brain-computer interfaces allow people to manage electronic devices such as computers without using their motor nervous system. When the brain is in a function, nerve cells in the brain communicate with each other with electrochemical interactions. Electroencephalogram (EEG) signals are recorded with the aid of electrodes during this function of the brain. These signals enable interaction between people and electronic devices. This interaction forms the basis of brain computer interface (BCI) systems which facilitates lives of paralyzed patients who do not have any problems with their cognitive functioning. Therefore, for high-performance BCI systems, pre-processing technique and classification method applied to these signals and features extracted from these signals are crucial. In this study, we studied a new EEG data set recorded from 29 people during imagination of hand opening/closing movement. While moving average filter was used a pre-processing technique, the features were extracted by Hilbert Transform and Mean Derivative. Afterwards, extracted features were classified by k-nearest neighbor method. Average classification accuracy (CA) with pre-processing was achieved 82.23%, which was 12.78% higher than the average CA obtained by unprocessed EEG data set and 16.63% greater than the previous works reported in the literature. The achieved results showed that the proposed method has a great potential to be applied general with a highperformance in general.